Details
-
Type:
Epic
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: ap_association
-
Labels:
-
Epic Name:f18-ap-association
-
Story Points:50
-
WBS:02C.03.02
-
Team:Alert Production
-
Cycle:Fall 2018
Description
Improve the prototype source association pipeline. This will include:
- Integration with the l1dbproto provided by the DAX team, iterating with them on design and requirements, to result in a documented interface to which both teams can work;
- Generate simple time series statistics, to be agreed agreed with Eric Bellm, during source association (per LDM-151 v4.1 §6.21; see also DM-14437);
- Think hard about the edge cases in AP source association, and come up with potential solutions for implementation in the next cycle (per LDM-151 v4.1 §6.23.4).
Attachments
Issue Links
- relates to
-
- Done
| Key | Summary | Story Points | Assignee | Status | |
|---|---|---|---|---|---|
|
|
4 | Chris Morrison [X] (Inactive) | Done | ||
|
|
4 | Chris Morrison [X] (Inactive) | Done | ||
|
|
12 | Chris Morrison [X] (Inactive) | Done | ||
|
|
1 | Chris Morrison [X] (Inactive) | Done | ||
|
|
6 | Chris Morrison [X] (Inactive) | Done | ||
|
|
8 | Chris Morrison [X] (Inactive) | Done | ||
|
|
2 | Chris Morrison [X] (Inactive) | Done | ||
|
|
6 | Chris Morrison [X] (Inactive) | Won't Fix | ||
|
|
Write the formatter for serializing DIAObjects/Sources through the butler |
6 | Chris Morrison [X] (Inactive) | Won't Fix |
Activity
Note to self: this epic was originally scheduled for the S17 cycle. However, it was deprioritised due to the emphasis on the AP end-to-end system, for which a "minimal" association system is being developed in DM-10768. That means we're punting it to the future. That means we'll show a variance on it, but it is nicely explicable.
Closing out this epic and the end of F18. Changing priorities and requirements have caused this work to focus on integration of the PPDB with the AP system; time series statistics and “probablistic” source association will be addressed in DM-16720 during S19.
This should include Chris Morrison [X] working with Eric Bellm to define the set of functionality needed. We also need to define metrics that can be used to test association as well as the dataset(s) to feed the metrics.
This may require implementing
DM-8837.Other open questions:
What about matching to SSObject?
How do we assign a confidence to the association?
Can DIAObjects contain more than one DIASource from a given epoch?
Can DIASources belong to multiple DIAObjects?